DOES MANDATORY DISCLOSURE OF DIRECTORS’

AND OFFICERS’ LIABILITY INSURANCE CURB

MANAGERIAL OPPORTUNISM? EVIDENCE FROM

THE CANADIAN SECONDARY MARKET

NARJESS BOUBAKRI* AND NABIL GHALLEB**

September 2008

JEL Classification: G22, G32

Keywords: Directors’ and Officers’ liability insurance, managerial opportunism, seasoned equity

offerings, post-issue stock price underperformance

* Department of Finance, HEC Montreal, ** King Fahad University of Petroleum & Minerals. We are grateful to Sandra Betton, Martin boyer, Jean Bédard, Georges Dionne, Hatem Ghouma, and Oumar Sy for helpful comments. Mohamed Jabir provided useful research assistance. Our research has benefited from comments by participants at the 29th McMaster World Congress in Hamilton. Boubakri acknowledges financial support from the SSHRC while Ghalleb acknowledges financial support from CREF and Canada Risk Management Chair. Please address all correspondence to Narjess Boubakri, Department of Finance, HEC Montreal, 3000 Chemin de la Cote-Sainte-Catherine, Montreal (PQ) H3T 2A7, Canada, Email: narjess.boubakri@hec.ca.

2

DOES MANDATORY DISCLOSURE OF DIRECTORS’ AND

OFFICERS’ LIABILITY INSURANCE CURB MANAGERIAL

OPPORTUNISM? EVIDENCE FROM THE CANADIAN

SECONDARY MARKET

ABSTRACT

We test the managerial opportunism hypothesis in a new context: The purchase of D&O insurance

around the time when Canadian firms are getting ready to proceed with their seasoned equity

offerings (SEOs). The Canadian context is interesting mainly because the reporting of D&O

insurance details is mandatory. We examine whether informed managers modify their behavior

regarding the purchase of D&O insurance when their firm are about to sell SEOs. After controlling

for other cross-sectional determinants of D&O insurance coverage, we find a significant negative

relation between the amount of D&O insurance purchased and subsequent firm performance. This

result is surprising because the imposition of a public disclosure is supposed to make the D&O

insurance purchases a useless tool for extracting private benefits from private information. As a

further test of managerial opportunism, we investigate the premiums paid around the

announcement date. If the insurers are able to distinguish managerial opportunism from genuine

coverage, they will be able to charge larger insurance premiums on managers that try to cover

themselves from the misfortunes of issuing overvalued stocks. The empirical evidence shows that

D&O insurers are unable (or do not attempt) to charge more expensive premiums to firms that

exploit private information or that abnormally increase their insurance coverage for whatever

reason. Finally, we do not find any support to the prediction that lower D&O insurance premiums

are associated with good corporate governance quality. One major implication of this essay is that

the imposition of a mandatory reporting does not alleviate the opportunistic behavior of managers

through the purchase of D&O insurance.

3

Proponents for a liberal policy of corporate insurance argue that directors’ and officers’

(hereafter D&O) liability insurance alleviates agency problems by aligning management interests

with those of shareholders.1 However, there is mounting evidence that, by weakening the efficiency

of shareholder lawsuits as a managerial control device, D&O insurance favors opportunistic

behavior by managers who become more likely to increase their private benefits at the expense of

the shareholders.2

The managerial opportunism hypothesis suggests that, in a world of asymmetric information,

managers that possess superior information about the firm future performance will have an

incentive to issue equity when their firm is overvalued. Charmers, Dann, and Harford (CDH, 2002)

examine whether managers act opportunistically regarding the purchase of D&O insurance around

the time of the firm’s initial public offerings (IPO). If managers behave opportunistically when

issuing overvalued stocks, it is likely that they will try to hedge the individual risk resulting from the

exploitation of their private information. Given that lawsuits are typically triggered by unfavorable

news in the aftermarket, the likelihood of being sued by unsatisfied shareholders after a sudden

plunge in market price is one of the most important costs of issuing overvalued stocks. In such a

case, D&O insurance will protect managers against the risk of shareholder litigation. Evidence in

CDH (2000) supports the hypothesis that managers behave opportunistically during the IPO

process, as the authors find a negative association between the purchase of D&O insurance

purchased and post-IPO stock price performance.

1 According to this view, D&O insurance is important because it can improve the monitoring role of public companies’ shareholders by attracting and maintaining competent directors and officers (Priest (1987), Daniels and Hutton (1993), and O’Sullivan (1997, 2002)). D&O insurance can also counter managers’ natural tendency to be overly conservative (risk aversion) in managing their firm when they are not enough protected against liability risk (Jensen (1993)). D&O insurers prevent losses by offering an incomplete protection to directors and officers and leave part of the risk to be borne by the corporation, via deductibles and limited coverage (Baker and Griffith (2006)). Finally, litigation can still be an important control device even if all direct costs are paid by an external insurer, if there are reputation costs associated with losing lawsuits (Bhagat, Brickley, and Coles (1987)). 2 If insured managers expect greater immunity from liability in suits by shareholders, the potential for lawsuits may do little to discipline them into working in the shareholders’ interests (Core (1997, 2000)). Furthermore, if outside directors believe that managers are exempted from liability in suits, they will likely reduce their (unobservable) effort in monitoring their executive colleagues, because of a lower probability to discipline them successfully (Core (1997) and O’Sullivan (2002)). Exposing outside directors to a lower risk level through D&O insurance is also likely to dampen their efforts in monitoring managers.

4

In this study, we test the managerial opportunism hypothesis in a new context: The purchase

of D&O insurance around the time when a firm is getting ready to proceed with its seasoned equity

offerings (SEOs). We examine whether informed managers modify their behavior with respect to the

purchase of D&O insurance when their firm are about to issue SEOs. After controlling for other

cross-sectional determinants of D&O insurance coverage, we find a significant negative relation

between the amount of D&O insurance purchased and subsequent firm performance. As a further

test of managerial opportunism, we examine the premiums paid around the announcement date.

Given the structure of their incentives and their unique access to information, one can expect

insurers to develop expertise in differentiating good and bad risks and to build these assessments

into their models for pricing D&O insurance. In other words, if the insurers are able to distinguish

between managerial opportunism and genuine coverage, they will be able to charge larger insurance

premiums on managers that try to cover themselves from the misfortunes of issuing overvalued

stocks. Because the insurance covers fully their private risk while the cost is spread among all

shareholders, managers mismanaging private information will be willing to pay abnormally high

premiums. Hence, one can expect a negative relation between D&O insurance premium and private

information, proxied by abnormal future performance. Our primary evidence does not support this

conjecture and shows that D&O insurers do not (or do not attempt) to charge higher premiums to

firms that exploit private information or to those that abnormally increase their insurance coverage

for whatever reason. However, when we use a three-stage least squares (3SLS) model to control for

the potential simultaneous determination between D&O insurance coverage and premium, we find

results that support our expectation of a negative relation between D&O insurance premium and

abnormal future performance.

Our work extends the literature in two main directions. (a) Firstly, while the existing studies are

based on non-mandatory U.S. information obtained from private sources, this study focuses on

Canadian proxy statement data. Because the reporting of D&O insurance details is mandatory in

Canada, our sample is free of the sample selection problems associated with non-mandatory data.

Moreover, the importance of studying managerial opportunism in a context where disclosure is

5

mandatory has been highlighted respectively by CDH (2002, p.633) and O’Sullivan (2002, p. 582) in

the following terms:

“In the context of D&O liability insurance, it would seem that if managerial

opportunism has an impact on the directors’ and officers’ insurance decision, then the

merit of mandatory disclosure of directors’ and officers’ liability insurance should be

studied further. We note that some countries, such as Canada, have chosen to require

full disclosure of these decisions.’’

“Future research could also examine the ex-post behavior of insured companies in

order to identify whether the purchase of D&O insurance represents any element of

managerial opportunism. This work is useful in informing policy deliberations in the

UK and elsewhere concerning the need for companies to disclose information on

director indemnification provisions.’’

One expected benefit of the mandatory reporting of D&O insurance details is its ability to curb any

managerial opportunism behavior associated with the purchase of insurance.3 However, even within

such a mandatory framework (i.e. the Canadian market), we find that the D&O insurance decision

reveals significant managerial opportunism, suggesting that, in fact, mandatory reporting doesn’t

fully alleviate managerial opportunism.

(b) Secondly, we test the managerial opportunism in the secondary market (SEO issuances)

rather than in the primary market (IPO issuances). We consider that this is a more appropriate

exercise because we can examine the managerial behavior change with respect to D&O insurance

purchase decisions around the event date—prior to the IPO date, such information on D&O

insurance is not publicly. Moreover, since SEO issuers are already followed by analysts, offer public

and audited information, have greater market capitalization, and are easier to sell short, it seems

3 CDH (2002) argue that this mandatory disclosure of D&O insurance details is more likely to limit potential managerial opportunism associated with this purchase decision while Griffith (2005) argues that the basic benefit of this disclosure is improvement of capital market efficiency through the signaling effects provided by D&O insurance policy details.

6

plausible to conjecture that incentives and opportunities to act opportunistically are more limited

for SEOs than for IPOs. Thus, any evidence of predictable post-issue returns for SEO issuers

based on available D&O insurance information poses an even stronger challenge to the efficient

markets theory than the evidence on IPOs. Using univariate tests, we document that managers do

significantly change their D&O insurance purchase around the equity offering dates. This suggests

that managers time their purchase of insurance to coincide with their equity-offering surge. This

makes sense only when there is opportunistic behavior. The relation between D&O insurance and

post-issue stock price underperformance has also been documented by CDH (2002) for a sample of

IPO firms.

The remainder of this study is organized as follows. Section I describes the basic features of

D&O insurance. Section II briefly reviews the literature and develops the main hypotheses. Section

III presents the data and summary descriptive statistics. Section IV presents the empirical results,

while Section V concludes.

I. Directors’ and Officers’ Liability Insurance in Canada

In recent years, D&O insurance has grown to become a core component of corporate insurance.

This growth has been stimulated by the fact that it has become routine that disgruntled investors

accuse corporations and their D&O with securities fraud whenever a firm’s stock price declines

dramatically and unexpectedly. In this section, we describe and discuss the D&O insurance and the

difference between Canada and the U.S.

A typical D&O insurance policy, which is a group policy purchased by the corporation,

includes three basic types of coverage: First, a personal coverage protects each individual officer or

director against the risk of shareholder litigation. Under this coverage, the insurer will pay covered

losses on behalf of managers when the corporation is not able or unwilling to indemnify its

managers by reason of law or because of financial distress. Second, a corporate coverage protects

the corporation itself against losses incurred from its indemnification commitment to individual

7

D&O. Third, an entity coverage protects the corporation itself when this latter is a defendant in a

shareholder claim.

D&O insurance policies cover losses including damages, judgments, awards, settlements

amounts and defense fees incurred in a shareholder claim. Moreover, the insurer covers directors

and officers for a claim alleging wrongdoing4 in their capacity as D&O, provided they acted

honestly and in good faith5 and with a view to the best interests of the corporation. Actions

qualified as fraudulent, illegal or involving evident conflicts of interest are typically excluded from

D&O insurance coverage.6 In general, D&O insurance coverage is stated on an annual basis for all

covered losses during the policy year. The personal and corporate coverage are usually the same.

The lawsuit risk is the greatest risk facing directors and officers, and has become more acute

after the recent corporate scandals, which involved quite reputable firms such as Enron, Health

South, Tyco, and World Com in U.S. and Nortel in Canada. For public firms, the dominant source

of D&O risk is shareholder litigation. According to Tillinghast-Towers Perrin’s (2003) Directors

and Officers Liability Survey, about one-half of D&O insurance claims are brought by shareholders

in both Canada and the U.S. (the remainder claims are triggered by employees, customers,

competitors, regulators, and other third-parties). Several events may trigger lawsuits filing: Illegal

acts, violations of security laws, self-dealing, control of the firm, unfair transactions, imprudent or

negligent management, bankruptcy, executive compensation, earnings restatement announcement,

and improper decision-making process and other self-interested transactions. There is a little cost to

filing, and several plaintiffs’ attorneys may file a suit. The filing includes an estimate of damages,

which is often based on the drop in market price prior to filing.

4 A simple definition of a wrongful act given by most professionals would be any error, misstatement, act, omission, neglect, or breach of duty committed, attempted, or allegedly committed or attempted, by an insured, individually or otherwise, in their insured capacity. 5 Effort made, information given, or transaction done, honestly and without a deliberate intention to defraud the other party. However, good-faith does not necessarily mean absence of negligence. 6 See Baker and Griffith (2007) for a detailed discussion on the principal D&O insurance exclusions.

8

Our interest in the Canadian data stems for several reasons: First, unlike their counterparts in

the U.S., Canadian securities regulators do require disclosure of basic information concerning their

D&O insurance policies, including coverage limits and premiums in their proxy filings and

registration statements. Therefore, Canada provides us with a natural laboratory to examine issues

related to D&O insurance as D&O insurance purchases must be publicly disclosed and are thus

supposed to be less contaminated by opportunistic behavior of directors and officers. Second, a

major difference between the U.S. and Canada pertains to the legal system: Litigation in the U.S. is

considered as a normal business expense by some firms (Kaltchev (2004)) contrary to the Canadian

market where litigation is less common. Core (2000) finds that the Canadian legal system is less

conducive to nuisance suits over stock price declines, probably because it is a considerably less

favorable environment for entrepreneurial plaintiffs’ lawyers (Baker and Griffith (2007)).

II. Literature Review and Hypothesis Development

Our study bridges the gap between D&O insurance and managerial opportunism in the secondary

market. Both topics have been examined in earlier studies. However, while the literature on

managerial opportunism in the equity offering context is large, that on D&O insurance liability is

quite limited.

So far, the existing D&O insurance studies focus primarily on the demand of insurance. Core

(1997) pioneered the study of the determinants of D&O insurance in Canada by showing that

litigation and distress risks are the major determinants of D&O insurance purchases in Canada.

O’Sullivan (2002) investigates the determinants of U.K firms' demand for D&O insurance and finds

that insured companies are larger, are more exposed to US litigation, experience greater share price

variability, exhibit lower levels of managerial ownership, and possess greater non-executive

representation on their boards than uninsured companies. Boyer (2003) shows that the lagged

insurance purchases is the only determinant that appears to be significant in explaining the

insurance decision in Canada, suggesting that habit is the main driver of D&O insurance demand.

9

Kaltchev (2004) instead argues that market capitalization is the main predictor of D&O insurance

coverage purchased by U.S firms. Holderness (1990) takes a corporate governance perspective and

shows that directors' and officers' (D&O) insurance has an important governance role in publicly

owned companies. Using a sample of U.K firms, O’Sullivan (1997) tests Holderness's monitoring

hypothesis by examining the association between board composition, managerial ownership,

external shareholder control, and the purchase of D&O. The results generally support the

monitoring hypothesis. Fewer studies focused on D&O insurance premiums, and the relation

between insurance purchase and firm value. For instance, Core (2000) tests whether D&O

insurance premium is commensurate with the firms’ corporate governance practices and finds that

it does. Bhagat, Brickley, and Coles (1987) examine stock price performance around the

announcement of the purchase of D&O insurance, and find no evidence that D&O insurance

purchase adversely affects shareholders’ wealth. Boyer (2005) provides evidence that D&O

insurance protects shareholders’ wealth rather than the directors’ welfare.

The literature related to managerial opportunism (market timing behavior) in the secondary

market is extensive. Indeed, several studies examining the firm's equity offering decision as an

alternative way to raise additional financing is based on the hypothesis that there is informational

asymmetry between insiders and outside investors regarding the true value of the firm, and posits a

decline in stock prices after the SEO. For example, Asquith and Mullins (1986) and Masulis and

Korwar (1986) document a significant price drop of the order of 2-3 percent upon the

announcement of SEOs. Eckbo and Marsulis (1992) also find an average 3 percent decrease in

prices at the SEO announcement dates. This drop is attributed to the investors revising downward

their beliefs about the value of the firm's stock. Ritter (1991) argues that managers issue equity

when their firms are overvalued, which would explain both the poor post-issue performance, and

the concentration of equity issues in periods of high market valuations. More recently, Loughran

and Ritter (1995) and Spiess and Affleck-Grave (1995) show that, on average, issuing firms

experience substantial and significant stock price underperformance over an extended period of five

years following their SEOs. They suggest that it is the revision of investors' optimistic expectations

10

of future performance that causes the poor long-run performance post-SEO. Loughran, Ritter, and

Rydqvist (1994) demonstrate that firms tend to issue at market peaks. Cheng (1994) argues, that

overvalued firms have incentives to issue equity even in the absence of good projects reaching the

conclusion that firms that do not invest the proceeds may be timing their issues in periods when

overpricing is most severe. Bayless and Chaplinsky (1996) investigate whether there is a window of

opportunity for equity issuers, and find that offerings indeed cluster when information asymmetry is

least severe. Finally, Teoh, Welch, and Wong (1998) and Rangan (1998) examine earnings

management in anticipation of SEOs. They find that reported earnings increase abnormally prior to

seasoned equity offerings, and show that issuers who alter their discretionary current accruals to

report higher net income prior to the offering will actually experience a larger post-issue negative

drift.

The only study that bridges the gap between D&O insurance purchases and managerial

opportunism in an equity offering context is CDH (2002). Using a sample of 72 U.S. firms, CDH

examine the decision of managers to purchase D&O insurance around the time when a firm is

preparing to sell its IPO. They hypothesize that—when managers opportunistically exploit their

private information—the amount of insurance coverage will be inversely related to the post-

offering performance of the issuing firm’s shares. They find a significant negative relation between

D&O insurance purchase and long-run future performance, suggesting the presence of managerial

opportunism. CDH (2002) also examine the relation between insurance premiums and future

performance and find that insurers do not (or are unable) to differentiate between purchasers with

depressing private information about the firm prospects and those that purchase extra coverage for

other reasons.

However, CDH’s (2002) study is based on privately obtained data, and it is not known whether

and how mandatory disclosure would alter the association between D&O insurance purchase and

managerial opportunism. As highlighted by CDH, one of the possible effects of the mandatory

reporting of D&O insurance details is its potential to limit managerial opportunism associated with

11

the purchase of insurance, mainly because of the release of private information associated with

mandatory reporting. Therefore, it becomes particularly interesting to investigate managerial

opportunism in the Canadian context where reporting is mandatory. We hypothesize that post-SEO

stock price performance will not be related to the D&O insurance amount purchased.

H.1: The post-SEO stock price performance will not be related to the D&O insurance

amount purchased.

Given the structure of their incentives as well as their unique access to information, one can

presume insurers will develop the ability to differentiate good and bad risks and to incorporate

these assessments into their pricing of D&O insurance.7 In other words, if insurers are able to

distinguish managerial opportunism from genuine coverage, they will likely charge larger insurance

premiums on managers that try to cover themselves from the misfortunes of issuing overvalued

stocks. Hence, we hypothesize a negative relation between D&O insurance premium and private

information, which we capture with the firm’s abnormal future performance.

H.2: The D&O insurance premium is inversely related to the private information

measured with the firm’s abnormal future performance.

III. Data and Summary Descriptive Statistics

A. Sample Selection

As previously discussed, we focus on Canadian firms because, unlike their U.S. counterparts, they

must disclose information regarding their D&O insurance policies, including coverage limits and

premiums in their proxy filings and registration statements. We have obtained an initial sample of

7 D&O insurers must develop expertise in assessing their potential payment obligations to be able to charge appropriate premiums for their risks because the survival and success of their business depend primarily upon taking in more capital than it pays out. To achieve this purpose, the pricing process of the D&O insurance premium is structured to enable the insurer to obtain cheap and full information about the firm’s litigation risk (Holderness (1990) and Knepper and Bailey (1993)).

12

SEOs from the Securities Data Corporation (SDC) database from January 1997 to December

2003.8

We imposed that all issues be Canadian ordinary common stocks; dropped all warrants, unit

issues, closed-end funds, right issues, and private placements; excluded financial firms (e.g.,

Loughran and Ritter (1995));9 excluded firms that are not covered by COMPUSTAT at the time of

the offering. To avoid statistical problems associated with interdependence of observations, we

include only the earliest offering if the sample firm has more than one SEO during the sample

period. Applying these requirements, we are left with 249 SEOs from 1997 to 2003.

We merged these 249 SEOs with our D&O insurance liability insurance source of data, the

SEDAR online database (www.sedar.com). The D&O insurance data include details on D&O

insurance amounts, premiums paid, and corporate deductibles. About 91 SEOs are dropped

because we are unable to access to important appropriate documents (proxy statement and/or

prospectus). Six SEOs are dropped because their D&O insurance coverage is not disclosed

although firm claim the purchase of coverage. One SEO is lost because D&O insurance coverage is

given by the firm’s parent and eight SEOs are dropped because the premium information is not

revealed although details on D&O insurance amount purchased are disclosed. Finally, five SEOs

are excluded because the firms state their intention to purchase D&O insurance but finally did not

carry it or did not disclose any related details. With these filters, we end up with 138 SEOs, out of

which 112 carry D&O insurance (81.2 percent) and 26 SEOs do not. 10

8 We start in 1997 because our main source of D&O insurance data (SEDAR) becomes available from this year. SEDAR is an online database that provides access to most public securities documents and information filed by public companies and investment funds with the Canadian Securities Administrators (CSA) in the SEDAR filing system. Our sample period ends in 2003 in order to ensure at least three years of post-SEO returns. 9 We remove financials since these firms’ incentive to issue equity (for example, banks may issue equity to meet regulatory capital requirements) as well as the applicability of multiple based valuations to these firms are not comparable to other firms. However, following Brav, Geczy, and Gompers (2000) and Eckbo, Masulis, and Norli (2000), we include SEOs of utilities in our sample. Exclusion of utilities from the sample does not change our results materially because the proportion of utilities is very small, less than 3 percent. 10 We ascertain whether our sample firms carry D&O insurance policy by examining their proxy statements and annual reports. If there is no mention of the existence of D&O insurance policy or no hint that the SEO firm has intention to purchase such policy, we consider the firm as uninsured.

13

B. Estimation of Private Information

Because shareholder lawsuits can be triggered if the subsequent aftermarket price falls below the

offering price, it is important to use a good and realistic long-run performance measure, that

typically captures the management’ private information. In this study, we choose to use buy-and-

hold returns (hereafter BHRs) as a proxy of long run performance.11 BHRs have become standard

in measuring long-term abnormal returns (see Barber and Lyon (1997) and Lyon, Barber, and Tsai

(1999)). BHRs measure the average multi-year return from a strategy of investing in all firms that

complete an event and selling at the end of a pre-specified holding period, versus a comparable

strategy using otherwise similar non-event firms. Even though they tend to have a highly skewed

distribution, BHRs are better than other proxies as CARs because the return differences are

obtainable by a realizable investment strategy, representing the actual wealth effect for an investor

in a more realistic way.

Following Loughran and Ritter (1995) and Spiess and Affleck-Graves (1995), we use three-year

BHRs to measure long-term performance. In particular, for each issuing firm, we calculate the

BHRs by compounding daily returns over either 756 trading days or the number of trading days

from the offering date until the delisting date, whichever is smaller. We use the same holding

periods and the same procedure to calculate the BHRs of matching firms. Finally, we compute the

buy-and-hold abnormal returns (BHARs) as the difference between the SEO firm buy-and-hold

return and the portfolio/firm buy-and-hold return. The three-year BHARs for sample firm i are

measured using the following formula:

[ ][ ]T delist T delist

i it benchmarktt first trading day t first trading day

BHAR R Rmin , min ,

(1 ) (1 )= =

= + − +∏ ∏ (1)

11 An alternative method of computing long-run stock price performance is to use cumulated abnormal returns (CARs). The difference between CARs and BHARs is that CARs ignore the effects of compounding returns. While the use of CARS has the advantage of easier statistical tests, it is hard to interpret the results in a practically meaningful way. Following Barber and Lyon (1997) and Kothari and Warner (1997), we do not use CARs in our study.

14

where ,i tR is the return of the sample firm i in day t , benchmark tR , is the return of the benchmark

over the same period, T is equal to 756 business days and delist is the delisting date, if any, for the

sample SEO firm. Following CDH (2002), we exclude the first day return. Three types of

benchmark are used depending on different matching samples strategies: Size, size and book-to-

market (B/M), and value-weighted market index.12

C. Variable Description

As dependent variables, we use D&O insurance coverage purchased (coverage) after the SEO

announcement and premium paid (premium). All variables are defined in Table II.1. We use three

categories of independent variables, namely: Firm characteristics, business risk variables, and

corporate governance variables. We control for two issue characteristics: The offering size and the

fraction of firm sold.

[INSERT TABLE II.1 ABOUT HERE]

We follow the literature and rely on five business-risk variable. Because the firm financing

policy is likely to affect the demand of corporate insurance, we control for the firm leverage.

Indeed, since distressed firms tend to exhibit a higher litigation risk and larger bankruptcy costs,

leverage is a potential determinant for D&O insurance demand. Therefore, we expect that more

leveraged firms are more likely to carry higher limits of D&O insurance and pay higher premiums.

Mayers and Smith (1982) argue that it would be optimal for firms that are large, healthy, mature,

and unregulated to purchase less insurance, mainly because they have more stable and confident

relations with different market participants and are less exposed to litigation risk than young firms.

Therefore, we hypothesize that age will be inversely proportional to D&O insurance coverage

limits and premiums. Since the firm’s past performance may be a reliable signal of its future

performance, we assume that litigation risk is negatively associated to the firm’s past revenues.

12 See Loughran and Ritter (1995) (for size) and Brav and Gompers (1997) (size and B/M).

15

Consequently, we expect to find a negative association between coverage and premium on the one

hand, and the firm’s past revenues on the other. Because risky firms are more exposed to litigation

risk, we expect a positive relation between standard deviation of revenues and D&O insurance

coverage and premium.13 Since liability risks differ between the U.S. and Canada—because of

differences in their respective legal systems—we use information on U.S. exchange listing firms

to test whether there is a different demand for D&O insurance and premiums in Canada. Canadian

firms listed on a U.S. exchange are exposed to a higher litigation risk as shown by Core (1997,

2000), carry higher D&O insurance limits and pay more expensive premiums for insurers. Thus, we

expect U.S. exchange listed firms to have a greater demand for D&O insurance and pay higher

premiums.

To examine how the governance structure of the firm affects D&O insurance, we rely on four

corporate governance variables: (a) Firms with a large board size are more likely to purchase higher

amounts of D&O insurance and pay higher premiums, as larger boards may increase agency

problems (see Yermack (1996)). An alternative hypothesis could be that larger boards may bring

together specialists from various functional areas, which will, in turn, increase firm value (Beiner et

al. (2006)). Under this alternative hypothesis, firms with large boards will have less need to carry

higher D&O insurance coverage limits and to pay higher premiums. The purchase of D&O

insurance can improve the monitoring role of public companies’ shareholders (e.g. Holderness

(1990), Daniel and Hutton (1993), and O’Sullivan (1997)). (b) Outside directors can be considered

as an alternative monitoring mechanism, we expect the amount of D&O insurance coverage and

premium to be negatively related to the number of outside directors. A high D&O ownership will

work on mitigating the agency problems, and we expect it to alleviate managerial opportunism as

well as D&O insurance needs. (c) The percentage of common shares held by blockholders can

help shareholders to effectively monitor the firm’s managers (Schleifer and Vishny (1986)). Hence,

we expect firms with blockholders to exhibit less need for D&O insurance. (d) D&O insurance is

13 We tested our models using average operating income and standard deviation of operating income instead of average revenue and standard deviation of revenue to find similar results.

16

often considered as part of the D&O compensation package (Core (1997)) and therefore can be

seen as a substitute to alternative forms of directors and officers compensation. Thus, we expect to

find a negative association between executive cash compensation and D&O insurance.

Finally, we investigate the role of auditors. Auditors, being the insurers of the firm’s financial

statements, mitigate the information asymmetry between inside managers and outside shareholders

(Klein (2002)). However, we expect managers of firms having one or more reputable auditors to be

exposed to higher litigation risk due to the rigorous monitoring generally conducted by these

auditors. Thus, firms with reputable auditors are hypothesized to carry higher D&O insurance

coverage because of managers’ risk aversion. Therefore, the net effect of this variable on D&O

insurance is ambiguous.

D. Summary Descriptive Statistics

Table II.2, Panel A, shows the distribution of our sample 112 SEOs by year. About 48 percent (54

firms) of our SEOs sample of SEOs occurs in the window 1999-2000. The distribution of the

sample by industry is shown in Panel B of Table II.2. Industries are identified by the Standard

Industrial Classification (SIC) codes provided by COMPUSTAT. Our sample is distributed across

11 industries. More than 50 percent (61 firms) of the sample firms are comprised in three sectors,

namely Biotechnology and Pharmaceuticals, Natural resources and Technology, and Mining.

Overall, the distribution of our sample across time and industry is consistent with the idea that

equity offerings occur in waves (see Lowry (2003), Rajan and Servaes (1997), and Pagano, Panetta,

and Zingales (1998)).

[INSERT TABLE II.2 ABOUT HERE]

Table II.3 presents summary descriptive statistics for the variables. Panel A of Table II.3 shows

the results for D&O insurance policy characteristics. The post-SEO mean D&O insurance coverage

and premium are roughly $19 million and $98,000, respectively. The average D&O insurance

17

coverage purchased is not puny, and represents on average about 87.5 percent of the SEO size.

Albeit not reported, the coverage represents about 8.9 percent of the market value of SEOs equity.

Interestingly, we record 59 firms (about 52 percent of our sample) that increase their D&O

insurance after the SEO. We find that 17 firms (15.3 percent of the sample) purchase a D&O

insurance policy for the first time after the SEO event. Our results are comparable with those

reported in the 2006 CIBC Director and Officer Liability Insurance Survey. According to the

survey, more than half of the firms that carry D&O insurance do so for less than $20 million

coverage, for less than $150,000 annual premiums, and less than $250.000 deductibles. The D&O

insurance become more costly for firms after SEOs, as the per dollar average cost of D&O

insurance coverage increases from 4.5 cents before SEOs to 5.4 cents after SEOs.

Panels B to D of Table II.3 shows the statistics for the other variables. The average age of

firms in our sample is 16 years. The average firm size in our sample is about $377 million, which is

substantially larger than the $20 million observed in CDH’s sample of IPOs. Our sample is also

characterized by high financial leverage. The median SEO offering size is $34 million and the

median market capitalization is $278 million. On average, the board is composed of 8.6 members.

The average fraction of outside directors is 67.8 percent. On average, directors and officers own

13.5 percent of their firm’s total shares outstanding after SEO, while blockholders own about 27.03

percent of the shares. Half of the sample (50.9 percent) has multiple SEOs (returns to the public

equity market at least once). About 10 percent of our sample purchases the D&O insurance policy

from an U.S. insurer and around 5 percent are listed on one of the U.S. stock exchange markets,

namely AMEX, NYSE, or NASDAQ. Finally, about 87 percent of our sample SEOs have one of

the big four auditors.

[INSERT TABLE II.3 ABOUT HERE]

18

IV. Empirical Results

We conduct our empirical analysis as follows. First, we estimate the likelihood of carrying a D&O

insurance policy using a probit model. Second, we perform a univariate analysis to investigate

whether managers do significantly change their D&O insurance purchase around the equity offering

dates. Third, as a more powerful test of managerial opportunism, we perform a multivariate analysis

to study the determinants of D&O insurance coverage and premiums paid by SEO firms. Finally,

we conduct several sensitivity tests to check the robustness of our results.

A. D&O Liability Insurance Purchase Decision and Long-Term Performance

To examine the relation between the post-SEO long-term performance and the likelihood of

carrying a D&O insurance policy, we run a probit model where the explanatory variables are

regressed on the dichotomous variable taking the value of one for insured firms and zero otherwise.

Table II.4 presents the results of probit regressions. The control variables used in these regressions

are the same as those we use in the next section to explain the total amount of coverage limit

purchased by our sample firms. Four different specifications are estimated: A first specification

estimates the basic model (before including the private information proxies) and the three other

specifications add a proxy of long-run future performance. We find a strong negative and

significant (at the 1 percent level) association between age and the likelihood to carry D&O

insurance. This result supports the prediction, ceteris paribus, that mature firms exhibit less needs

to purchase D&O coverage than young firms. Reported results also show that the fraction sold is

negatively and significantly related to the probability to purchase D&O insurance. Unexpectedly,

average revenue, and the standard deviation of revenues are statistically significant in most

specifications, but do not have the predicted sign. This result suggests that companies experiencing

larger revenues (more volatile revenues) are more (less) likely to possess insurance. We have no

explanation to this puzzle.

19

[INSERT TABLE II.4 ABOUT HERE]

The findings obtained for the long-run stock price performance of the SEO firms show a

negative association between this performance and the likelihood to carry D&O insurance policy.

However, this relation is not statistically significant at conventional levels for all specifications. The

limited size of the sample may contribute to the failure of our probit regressions to establish a clear

link between the probability to purchase D&O insurance and long-run performance. Nonetheless,

we are convinced that the lack of power in our probit model does not significantly weaken the

assumption that opportunistic managers are more likely to purchase higher levels of D&O

insurance than non-opportunistic ones.

B. Does D&O Liability Insurance Vary around SEO Announcements

One of the advantages of using an SEO sample is that we can track the managerial behavior change

with respect to D&O insurance purchase decisions around the event date.14 In this section, we

investigate the evolution of the D&O insurance coverage, premium, and deductible around the

SEO date. We compare the means and medians of these variables pre- and post-SEO. Table II.5

shows the results.

The post-SEO mean D&O insurance coverage and premium are $18.92 million and $100,822,

respectively. These figures represent significant increases (at the 1 percent level of statistical

significance) relative to the pre-SEO figures (respectively $13.93 million and $75,900). The average

corporate deductible significantly (at the 1 percent level) increases from $95,000 pre-offering to

$141,000 post-offering.

A similar difference between the pre- and post-SEO D&O insurance purchase obtains when

we consider the median differences. All Wilcoxon sum-rank tests show that the median differences

of coverage limit, premium paid, and corporate deductible around the SEO event are reliably

14 In contrast, information regarding D&O insurance is not publicly available prior to IPO date.

20

significant at the 1 percent level. Overall, the results suggest that managers time their purchase of

insurance to coincide with their equity-offering surge. This makes sense under the managerial

hypothesis, which is the focus of the multivariate tests presented below.

[INSERT TABLE II.5 ABOUT HERE]

C. D&O Insurance Limits and Long-Run Stock Price Performance

As hypothesized above, we expect that the post-SEO stock price performance will not be related to

the amount of D&O insurance coverage purchased, since the mandatory reporting of the purchase

of D&O insurance is equivalent to a private information release; managers would rather keep their

private information secret in order to fully profit from it. However, the univariate results presented

above point instead toward the presence of managerial opportunism, as both the D&O insurance

coverage and premium significance increases just after the SEO issuance.

In this section, we go beyond the univariate tests, and look at the multivariate determinants of

D&O insurance coverage. To control for the determinants of D&O insurance coverage under

normal circumstances, we build on Core (2000) and CDH (2002). Specifically, we consider a

regression model that combines business risk variables, corporate governance variables, and SEO

characteristics, and run the following regression model:

Coverage Risk Governance Characteristics e0 1 2 3 1α α α α= + + + + (2)

where Risk refers to the group of corporate business risk variables: The age of the firm, the firm’s

financial leverage, average revenues, the standard deviation of average revenues, and U.S. exchange

listing dummy; Governance stands for the vector of corporate governance variables: Board size,

percentage of outside directors on the board, D&O ownership, blockholder, executive cash

compensation, and big four auditor dummy; and Characteristics is the vector representing our SEO

characteristics: SEO offering size and fraction of firm sold.

21

Table II.6 shows the results from the estimation of the cross-sectional regressions based on

equation (2). Four different specifications are estimated: A first specification fit the basic model

from (equation (2)) and the three others add each a proxy of private information (long-run future

performance). Cross-sectional regression results show that leverage is positively and significantly

(almost at the 5 percent level) related to D&O insurance coverage. This result supports the

prediction that, ceteris paribus, higher leverage is associated with higher default risk. We find also a

strong positive and significant at the 1 percent level association between D&O insurance coverage

and the U.S. exchange listing dummy, suggesting that the firms listed on U.S. exchange markets

carry significantly higher D&O insurance coverage limits than firms listed on Canadian exchange

markets. This evidence is in line with our prediction that U.S. exchange listing exposes firms to a

greater probability of litigation risk, so much so that concerned managers have a great incentive to

purchase more coverage to protect themselves from shareholder lawsuits. Furthermore, firm age,

average revenue, and the standard deviation of revenues, although having the predicted sign, aren’t

distinguishable from zero at conventional levels of statistical significance.

[INSERT TABLE II.6 ABOUT HERE]

Of the corporate governance variables, only executive cash compensation and the big four

dummy are robustly and statistically reliable. The positive coefficient on the big-four auditor

dummy is consistent with the hypothesis that managers of firms reviewed by reputable auditors are

expected to carry higher coverage limits by fear of the rigorous auditing that can potentially create a

litigious environment, and cause shareholders lawsuits. The inferred positive relation between

executive cash compensation and D&O insurance coverage, however, contrasts with the

predictions of the efficient contracting models (e.g. Parry and Parry (1991)), as well as with the idea

that D&O insurance can be seen as a substitute to alternative forms of directors and officers

compensation.15 Furthermore, both SEO characteristics, namely offering size and fraction of the

15 The result is consistent with the assumption that various compensation forms complement each other. Core (1997, p.81) argues that this puzzling result is obtained because the compensation components “are

22

firm sold, are consistently significant at the 5 percent level with the expected signs, suggesting that

these characteristics variables are important determinant for D&O insurance coverage.

We obtain some interesting results on the long-run stock price performance of SEO firms,

which proxies for private information. For all specifications, we find a negative association between

D&O insurance coverage limit and the long-run stock price performance. This result is reliable at

the 1 percent level for the size-adjusted BHARs, at the 5 percent level for the market-adjusted

BHARs, and at the 1 percent level for the B/M and size-adjusted BHARs.

In unreported results, we follow CDH (2002) by including the standard deviation of daily

returns to proxy for the known risk of the firm prior to the SEO. As CDH (2002), we compute the

standard deviation of daily returns from day 10 to the end of the first year. This variable can also

capture the possibility that riskier SEOs will perform poorly in the future. The evidence shows that,

introduced alone or combined with one of our long-run performance proxies, the standard

deviation of daily returns fails to exhibit any significant effect on the amount of D&O insurance

coverage purchased by SEO firms. Therefore, we conclude that SEO firms exhibit a poor long-run

stock price performance not simply because they are riskier SEOs, but because their respective

managers were behaving opportunistically when issuing their equity on the market.

The evidence presented above does not support our hypothesis of absence of managerial

opportunism for Canadian SEOs. The negative and significant association between the amount of

D&O insurance purchased and different proxies of long-run stock price performance implies the

presence of managerial opportunism. Therefore, we can conclude that mandatory reporting is not

the panacea for the agency problems induced by managers’ private information. In other words, the

result suggests that available information on the D&O insurance purchase can be used to predict

future performance. Accordingly, the fact that managers act opportunistically despite the mandatory

reporting of their liability insurance details suggest that either: (a) Markets are inefficient since post-

bundled together by an insider who does not internalize their cost, or because little attention is given to designing director compensation plans’’. In other words, managerial entrenchment may prevent efficient contracting with outside directors.

23

SEO performance is predictable based on available D&O insurance information; (b) investors are

naïve or irrational because they extrapolate pre-issue earnings to systematically overestimate post-

issue earnings and ignore relevant public information contained in D&O insurance details; or (c)

obtaining and/or implementing the D&O insurance information are overly costly to the typical

investor despite the mandatory reporting. Even if it is not very costly to obtain a firm's D&O

insurance information from its financial statements, understanding the implications of this

information may be quite costly in terms of time, effort, and other investor resources (see for

example Indjejikian (1991)).

D. D&O Liability Insurance Premiums and Unexplained Coverage

In this section, we investigate the relation between D&O insurance premium and future

performance of Canadian SEOs. CDH (2002) do not find any significant effect between the

premiums paid and future performance. Following Core (2000), we investigate whether insurers can

detect managerial opportunism using a two-stage estimation procedure. The rationale behind the

use of this estimation method is that the amount of insurance premium is a function of the firm’s

business risk, its corporate governance, its offering characteristics and its D&O insurance limit

purchased.16 However, as shown by equation (2), the D&O insurance limit is, itself, a function of

business risk proxies and corporate governance. To resolve this problem, we first conduct a first-

stage regression by estimating equation (2) to obtain the unexplained coverage, which is the error

term from the regression. Then, in the second stage, we regress the premiums on a set of

independent variables including the residual coverage and the firm’s business risk, corporate

governance, and SEO characteristics variables. The two-stage regression can be easily understood

from the following equations:

Premium Risk Governance Characteristics Coverage e0 1 2 3 4 2β β β β β= + + + + + (3)

16 We consider the amount of D&O insurance coverage purchased as a determinant of premium. At this stage, we assume that the coverage limit is chosen first, and then the premium is settled following negotiations between the parties of insurance policy.

24

Substituting coverage in equation (3) by its expression in equation (2), yields

Premium Risk Governance Characteristics e e0 1 2 3 4 1 2γ γ γ γ γ= + + + + + (4)

where for each i =1,2,3,4 , we have i i i4γ β β α= + and 4 4γ β= .

The coefficient iγ represents the total effect of the independent variables on the premium.

Since the independent variables have an impact on both the D&O insurance coverage purchased

and the premium paid, we can decompose their total effect on the premium into a direct effect ( iβ )

and an indirect effect ( 4 iβ α ). The coefficient 4γ captures the effect of unexplained coverage on the

premium, making it possible to gauge the extent to which insurers are able to appraise the abnormal

insurance demand in their premium pricing process.17

Table II.7 shows the results for estimating several specifications based on equation (4). In the

first specification, we just estimate the equation without any alteration, while in specifications (2) to

(4) we use the interactions between the unexplained coverage and the long-run performance

proxies. These interaction terms are added to investigate whether insurers can distinguish between

insured SEOs that abnormally demand coverage because of negative private information from

those following the same behavior for other reasons. If insurers are able to detect opportunist

managers, we expect the coefficient on the interaction variables to be negative. Although having the

hypothesized sign across all our specifications, neither the unexplained coverage, nor the interaction

variables are significantly related to the premium. As found by CDH (2002) for a sample of U.S.

IPOs, this result suggest that insurers do not (or do not attempt) to charge more expensive

premiums to cover opportunistic managers issuing Canadian SEOs. In contrast to evidence in the

17 To estimate 1e , we run equation (2) while taking out all the independent variables that are not known at the time of the SEO.

25

U.S., D&O insurers are do not charge higher premiums even to firms that abnormally increase their

D&O insurance purchases for other reasons than opportunism.18

[INSERT TABLE II.7 ABOUT HERE]

Mindful of the differences between U.S. and Canadian legal systems, we believe that it is

difficult to generalize findings from U.S. (CDH (2002)) or any foreign jurisdiction to Canada.

Although unexpected, these results can somewhat be rationalized by the arguments discussed

below:

(a) It is common kowledge that the Canadian litigation environment is different from the

American. In the U.S., litigation is a normal business expense. As a result, insurers have stronger

incentives to appraise the shareholder litigation risk accurately for U.S. firms, compared to Canadian

firms. Therefore, the smooth nature of the litigation environment in Canada probably causes

insurers to underestimate the litigation risk to which Canadian firms are exposed.19

(b) An alternative explanation is related to market constraints. Insurers are naturally inclined to

sell insurance at larger premiums. However, this is not an easy task because of the competitive

constrains. If a particular insurer charges more than its market competitors for the same risk, she

will lose some market shares. Moreover, according to Tillinghast-Towers Perrin’s 2002 Directors

and Officers Liability Survey, about 75 percent of our sample firms issued SEO between 1997 and

2000 where the market was “soft’’ with decreasing premiums, increasing policy limits and

deductibles/retentions, and a rise of insurance capacity. The features of such market complicate not

18 Our results also show a significantly positive (negative) association between SEO size (fraction sold) and premium. Interestingly, we find a negative (positive) relation between average revenues (standard deviation revenues) and coverage premium. Since SEO size and fraction sold were significant determinants of coverage in the first stage (equation (2)), their significant effect on premium in the second-stage is at least coming from their indirect effect on D&O insurance limit. However, average revenues and standard deviation revenues were statistically insignificant in equation (2), hence their significant effect in equation (4) is totally due to their direct effect on premiums paid. 19 According to Tillinghast-Towers Perrin’s 2002 Directors and Officers Liability Survey, only 17 percent of Canadian participants reported one or more claims against their directors or officers over a ten-year experience period. In addition, the 2006 CIBC Director and Officer Liability Insurance Survey indicates that 84 percent Canadian companies do not report an actual claim or threatened claim in the previous three years.

26

only the insurers’ ability to adjust their risk-assessments to arrive at a competitive price, but also to

match premiums charged to the risks to be supported. Therefore, market constraints limit the

ability of insurers to incorporate all risk factors into the pricing of D&O insurance.20

(c) Unlike their U.S. counterparts, Canadian securities regulators do require disclosure of D&O

insurance details. Accordingly, one can plausibly believe—although this study gives opposite

evidence—that Canadian insurers may view the public nature of D&O insurance policy as a

deterrence device preventing, or at least minimizing, managerial opportunistic behavior via

manipulation of D&O insurance purchase (CDH (2002)). Canadian insurers are more likely to be

less diligent than U.S. insurers when pricing premium.

However, before advancing any conclusion or implication of this finding at this stage, it makes

sense to investigate whether our results are conducted by a potential simultaneity between D&O

insurance coverage and premium. The next section treats this issue.

E. Simultaneous Determination of D&O Insurance Limits and Premiums

In this section, we examine the determinants of D&O insurance coverage and premium

simultaneously. Boyer (2003, 2005) evokes the possibility that D&O insurance limit and deductible

are chosen simultaneously rather than independently. He argues that a risk-averse agent negotiates

simultaneously the D&O insurance limit and the deductible at the time of the insurance policy

purchase. It is plausible then to look at D&O coverage and premium as dependent rather than

independent decisions. Furthermore, these two D&O insurance policy pillars are related also

indirectly through their relationship with several control variables. Accordingly, using a standard

approach based on a single-equation technique can be misspecified, leading to spurious inferences.

20 See Baker and Griffith (2007) for a detailed discussion on constraints in the insurance market.

27

To control for this simultaneity relation between D&O insurance coverage and premium, we use a

three stage least squares approach (3SLS)21to estimate jointly equation (2) and equation (4).

Panel A, Table II.8, shows the results of estimating the coverage model (equation (2)) using the

3SLS approach. Overall, results are similar to those found using a single-equation OLS model.

Panel B of Table II.8, presents the 3SLS results of estimating the D&O insurance premium

(equation (4)). Recall that we investigate whether insurers can detect and price managerial

opportunism through abnormal coverage purchases. Contrary to the two-stage regression method,

we find a reliable significant and positive relation between unexplained coverage and D&O

insurance premium. Interestingly, the interaction variables between the unexplained coverage and

the long-run performance proxies are negatively and significantly related to the premium in most

specifications. This result suggests that insurers do not only pool all types of excess coverage

purchasers together, but they are also able to distinguish between firms that purchase abnormal

excess coverage by anticipation of poor future long-run performance and firms that purchase extra

coverage for some other reason, such as abnormal higher risk aversion by directors and officers.

[INSERT TABLE II.8 ABOUT HERE]

F. Additional Robustness Checks

In this section, we perform a number of additional robustness checks. We investigate the

robustness of our results in various contexts. Qian (2005) finds that investors react negatively to the

announcement of fast offerings. The author shows that the initiation of another equity offering

shortly after a previous one signals to market participants that the firm’s management believes their

stock is overvalued, and attempts to time the market. We thus test the managerial opportunism

hypothesis for firms issuing multiple SEOs, and we investigate whether growth firms are subject to

managerial opportunism. This hidden agency problem can explain why growth firms experiment

21 According to Belsley (1988), the 3SLS possesses greater small-sample efficiency than 2SLS, which also motivate our choice due to the small size of our sample.

28

post offering stock price underperformance (Lee (1997), Spiess and AfIleck-Grave (1995)) and

unanticipated deterioration in their earnings’ performance following their SEOs (Lee (1997)).

Table II.9, Panel A, presents the results of regressions that test managerial opportunism for the

firms issuing multiple SEOs. To do so, we replace in each specification of table II.5 the long-run

performance proxy by the product between the long-run performance proxy, and the multiple SEO

dummy, which takes 1 if the sample firm issues more than one SEO through the sample period and

0 otherwise. The coefficient associated with the multiple SEO interaction term is negative and

significant in most of the specifications, supporting the managerial opportunism hypothesis for

firms issuing frequent SEOs.

[INSERT TABLE II.9 ABOUT HERE]

Panel B of Table II.9 show the results of the estimation of the basic model specification

(equation (2)) augmented with the interaction terms between the long-run performance proxy and

the growth dummy, which takes 1 for firms having their market-to-book ratio is grater than 1 and 0

otherwise. The coefficient associated with this growth interaction term is a negative and significant

(at the 5 percent level for two of the five specifications. Taken together, in addition to the fact that

all estimated coefficient are consistently negative, our results support the managerial opportunism

hypothesis for growth firms.22

As we discussed above, we are convinced that our general model specification for both

coverage limit and premium were chosen to tightfistedly reflect, based on previous literature, the

pertinent factors representing firm’s business risk as well as its corporate governance quality. Even

so, we include numerous other variables in both coverage limit regressions and insurance premiums

specifications to ensure that omitted variables are not inducing spurious inference with respect to

the independent variables. More specifically, we have considered the average operating income and

22Because of the multicolinearity problem existing between the BHARs proxies and the interaction terms defined above respectively for multiple SEOs and growth firms, we were not able to examine whether these two sub-samples of firms exhibit a stronger managerial opportunism than the entire sample.

29

standard deviation of operating income as additional business risk variables; the average years on

the board of directors, the percentage of inside directors, D&O indebtedness toward firm and

institutional shareholding as further corporate governance factors, and the offer day return

(underpricing) as further SEO characteristics.23 However, none of our results and inferences is

materially affected by the inclusion of these variables.

The results are also robust to the inclusion of an alternative private information proxy, namely

the market equally-weighted index-adjusted BHARs. As expected, this proxy is negative and

significantly related to D&O insurance coverage purchased and premiums paid in almost all

regressions. Moreover, its inclusion does not alter our inferences. In addition, we rerun our tests

using natural logs in order to mitigate potential heteroskedasticity and the influence of a few

extreme observations. However, we find that the premiums paid and the average revenue are

negative for some firms. As an additional sensitivity test, we rerun our second-stage regression

(equation (4)) without interacting the long-run stock price performance proxies with unexplained

coverage. Overall, the results we obtain remain qualitatively the same.

Finally, we test for the bias that can result from using non-randomly selected samples. Using a

Heckman selection model to control for this potential bias, we find that our results remain robust.

Results are reported in Table II.10.

[INSERT TABLE II.10 ABOUT HERE]

V. Conclusion

This study examines the managerial opportunism hypothesis in the D&O insurance context in

Canada. The Canadian context is interesting mainly because the reporting of D&O insurance details

is mandatory. Our evidence supports the managerial opportunism hypothesis by showing that the

23 Although it is an important variable, we do not consider offer day return in our regressions because its inclusion reduces significantly our sample to 88 firms. We define offer day return as the return from the offering price to the end of the first trading day.

30

amount of D&O insurance is inversely related to the firm’s future performance. This result is

surprising because the imposition of public disclosure is supposed to make the D&O insurance

purchases a useless tool for extracting private benefits from private information. In any case, our

results show that the imposition of mandatory reporting is not the panacea for solving the agency

problems induced by managers’ private information. Moreover, the fact that managers act

opportunistically despite the mandatory reporting of their liability insurance details question the

semi-strong efficiency of the Canadian markets in reflecting fully the information available to all

investors.

Surprisingly, our findings do not support the prediction that insurers can distinguish

abnormally large insurance purchases driven by adverse information from those driven by abnormal

risk aversion. Indeed, D&O insurers of Canadian SEOs do not (or do not attempt) to charge more

expensive premiums to insured firms exploiting private information in their D&O insurance

demand, and higher premiums to firms that abnormally increase their D&O insurance purchases

for whichever reason. We do not find any support for the prediction that lower D&O insurance

premiums are associated with good corporate governance quality.

Our findings have important implications for investors, and regulators. Investors can learn

from firms’ D&O information to distinguish among issuers quality. For regulators, our evidence

provides some further insights on the debate on mandatory disclosure of D&O insurance that is

raging in the U.S. In fact, the mandatory disclosure of D&O insurance details in Canada fails to

limit potential managerial opportunism associated with the D&O insurance purchase decision. One

implication of this result is that if D&O insurance is more likely to have adverse incentive effects

that harm shareholders, then both Canadian and U.S regulators need to reconsider the

recommendation to limit such practice.

31

References

Alexander, J. C., 1991, Do the Merits Matter? A study of settlements in securities class actions,

Stanford Law Review 43, 497-598.

Asquith, P., and D. W. Mullins, 1986, Equity issues and offering dilution, Journal of Financial

Economics 15, 61-89.

Baker, T., and S. Griffith, 2006, The Missing monitor in corporate governance: The Directors'&

Officers' liability insurer, Georgetown Law Journal, forthcoming.

Baker, T., and S. Griffith, 2007, Predicting corporate governance risk: Evidence from the Directors'

and Officers' liability insurance market, University of Chicago Law Review 74, 1-57.

Barber, B. M., and J.D. Lyon, 1997, Detecting long-run abnormal stock returns: The empirical

power and specification of test-statistics, Journal of Financial Economics 43, 341-372.

Bayless, M., and S. Chaplinsky, 1996, Is there a window of opportunity for seasoned equity

issuance? Journal of Finance 51, 253-278.

Beiner, S., W. Drobetz, M. M. Schmid, and H. Zimmermann, 2006, An integrated framework of

corporate governance and firm valuation, European Financial Management 12, 249-283.

Bhagat, S., J. A. Brickley and J. L. Coles, 1987, Managerial indemnification and liability insurance:

The effect on shareholder wealth, Journal of Risk and Insurance 54, 721-736.

Boyer, M. M., 2003, Is the demand for corporate insurance a habit? Evidence from directors’ and

officers’ insurance, Working paper CIRANO 42.

Boyer, M. M., 2005, Directors’ and officers’ insurance and shareholder protection, Working paper,

HEC Montreal.

Brav, A., C. C. Geczy, and P. A. Gompers, 2000, Is the abnormal return following equity issuances

anomalous? Journal of Financial Economics 56, 209-249.

Chalmers, J. M. R., L. Y. Dann and J. Harford, 2002, Managerial opportunism? Evidence from

directors’ and officers’ insurance purchases, Journal of Finance 57, 609-639.

Cheng, L. L., 1994, Equity issue under-performance and the timing of security issues, Working

paper, Massachusetts Institute of Technology.

CIBC Centre for Corporate Governance and Risk management, 2006, Director and Officer liability

insurance survey, Simon Fraser University.

32

Core, J. E., 1997, On the corporate demand for directors’ and officers’ insurance, Journal of Risk and

Insurance 64, 63-87.

Core, J. E., 2000, The directors’ and officers’ insurance premium: An outside assessment of the

quality of corporate governance, Journal of Law, Economics and Organization 16, 449-477.

Daniels, R. J., and S. M. Hutton, 1993, The capricious cushion: The implications of the directors'

and officers' insurance liability crisis on Canadian corporate governance, Canadian Business Law

Journal 22, 182-230.

Dunbar, F., and P. Hinton, 1995, Analysis of shareholder damages claims and the potential impact

of legal reforms, Report, NERA Economic Consulting.

Eckbo, B. E., R. W. Masulis, 1992, Adverse selection and the rights offer paradox, Journal of Financial

Economics 32, 293-332.

Eckbo, B. E., R. W. Masulis, and O. Norli, 2000, Seasoned public offerings: Resolution of the new

issues puzzle, Journal of Financial Economics 56, 251–291.

Griffith, S., 2005, Unleashing a gatekeeper: Why the SEC should mandate disclosure of details

concerning Directors' & Officers' liability insurance policies, Working paper, University of

Pennsylvania.

Holderness, C. G., 1990, Liability insurers as corporate monitors, International Review of Law and

Economics 10, 115 -129.

Indjejikianr, J., 1991, The Impact of costly Information interpretation on firm disclosure decisions,

Journal of Accounting Research 29, 277-301.

Jensen, M. C., 1993, The modern industrial revolution, exit, and the failure of internal control

systems, Journal of Finance 43, 831-880.

Kaltchev, G., 2004, The demand for directors' and officers' liability insurance by US public

companies, Working paper, Southern Methodist University.

Kothari, S. P., and J. B. Warner, 1997, Measuring long-horizon security performance, Journal of

Financial Economics 43, 301-339.

Lee, H. W., 1997, Post offering earnings performance of firms that issue seasoned equity: The role

of growth opportunities, Quarterly Review of Economics and Finance 37, 97-114.

Lee, I., 1997, Do firms knowingly sell overvalued equity? Journal of Finance 52, 1439-1466.

Loughran, T., and J. R. Ritter, 1995, The new issues puzzle, Journal of Finance 50, 23-51.

33

Loughran, T., J. R. Ritter, and K. Rydqvist, 1994, Initial public offerings: International insights,

Pacific-Basin Finance Journal 2, 165–199.

Lowry, M., 2003, Why does IPO volume fluctuate so much? Journal of Financial Economics 67, 3–40.

Lowry, M., and G. W. Schwert, 2002, IPO market cycles: Bubbles or sequential learning? Journal of

Finance 57, 1171-1198.

Lyon, J., D. B. M. Barber, and C. L. Tsai, 1999, Improved methods for tests of long-run abnormal

stock returns, Journal of Finance 54, 165–201.

Masulis, R. W., and A. N. Korwar, 1986, Seasoned equity offerings: An empirical investigation,

Journal of Financial Economics 15, 91-118.

Mayers, D., and C. W. Smith, 1982, On the corporate demand for insurance, Journal of Business 55,

281-296.

O’Sullivan, N., 1997, Insuring the agents: The role of directors’ and officers’ insurance in corporate

governance, Journal of Risk and Insurance 64, 545-556.

O’Sullivan, N., 2002, The demand for directors’ and officers’ insurance by large UK companies,

European Management Journal 20, 574-583.

Pagano, M., F. Panetta, and L. Zingales, 1998, Why do companies go public? An empirical analysis,

Journal of Finance 53, 27–64.

Parry, M. E. and A. E. Parry, 1991, The purchase of insurance by a risk-neutral firm for a risk-averse

agent, Journal of Risk and Insurance 58, 30-46.

Priest, G. L., 1987, The current insurance crisis and modern tort law, Yale Law Review 96, 1521–

1590.

Qian, H., 2005, Timing of seasoned equity offerings: A duration analysis, Working paper,

Pennsylvania State University.

Rajan, R. G., and H. Servaes, 1997, Analyst following of initial public offerings, Journal of Finance 52,

507–530.

Rangan, S., 1998, Earnings management and the performance of seasoned equity offerings, Journal of

Financial Economics 50, 101-122.

Ritter, J. R., 1991, The long-run performance of initial public offerings, Journal of Finance 46, 3–27.

Shleifer, A., and R.W. Vishny, 1986, Large shareholders and corporate control, Journal of Political

Economy 94, 461-488.

34

Spiess, D. K., and J. Affleck-Graves, 1995, Underperformance in long-run stock returns following

seasoned equity offerings, Journal of Financial Economics 38, 243–267.

Teoh, S. H., I. Welch, and T. J. Wong, 1998, Earnings management and the post-issue

underperformance in seasoned equity offerings, Journal of Financial Economics 50, 63-99.

Tillinghast - Towers Perrin, 2001, 2002, 2003, Directors and officers liability survey: US and

Canadian results, http://www.towersperrin.com.

Yermack, D., 1996, Higher market valuation of companies with a small board of directors, Journal of

Financial Economics 40, 185-211.Alexander, J. C., 1991, Do the Merits Matter? A study of

settlements in securities class actions, Stanford Law Review 43, 497-598.

35

Table II. 1 Variable Description

This table describes the variables used. Panel A shows the dependent variables; Panel B shows the SEO Characteristic variables; Panel C describes the firm’s business risk variables; and Panel D shows the corporate governance variables. Panel E presents the long-term performance measures. For each variable, we define its measurement proxy, and we give the predicted sign of its effect on both D&O insurance limit and premium paid as well as its source of data. All figures are in Canadian dollar.

Variable Name Definition Predicted sign/Cov

Predicted sign/Prem

Source of data

Panel A. Dependent Variables

D&O coverage Total coverage limits purchased at the end of the SEO fiscal year ($ millions) + SEDAR

Premium Annual premiums paid by SEO firm to purchase its D&O liability insurance policy ($ millions)

SEDAR

Panel B. SEO Characteristics

SEO offering size Value of SEO ($ millions) ? ? SEDAR

Fraction of firm sold Ratio of SEO shares offered to total outstanding shares after offer ? ? SEDAR

Panel C. Firm’s Business Risk Variables

Leverage Total debt divided by the market value of equity plus book value of preferred plus book value of debt

+ + COMPUSTAT

Age Number of years of operating history provided in the prospectus - - SEDAR

Revenues Average reported revenues over three years preceding the SEO announcement date ($ millions). Partial years are converted to full years by taking the ratio of partial year revenues to prior year partial year revenues and multiplying by the prior full year revenue figure

- - SEDAR

Panel C. Firm’s Business Risk Variables (continued)

SD adjusted revenue Standard deviation of revenues ($ millions) over three years preceding the SEO announcement date

+ + SEDAR

U.S. exchange listing A dummy variable taking 1 if the firm is listed on the U.S. stock market (NYSE, NASDAQ, AMEX) and 0 otherwise

+ + SEDAR

Market-to-book The market-to-book ratio of assets + + COMPUSTAT

36

Table II. 1 - Continued

Variable Name Definition Predicted sign/Cov

Predicted sign/Prem

Source of data

Panel D. Corporate Governance Variables

Board size Number of members on the board of directors ? ? SEDAR

Pct. outside directors Percent of the total number of directors on the board that are outside directors ? ? SEDAR

Block holder The cumulative percentage of outstanding shares held by outside holders who own at least 10% of firm shares and who are neither officers nor directors

- - SEDAR

Executive cash compensation The sum of fixed compensation, annual bonuses and ordinary shareholdings for executives - - SEDAR

D&O ownership (%) Percent of total shares outstanding owned by the directors and officers of the firm ? ? SEDAR

Big 4 auditor A dummy variable taking 1 if the firm is audited by a big 4 auditor and 0 otherwise ? ? SEDAR

Panel E. Long-Run Performance Measures 3 years BHARs size & BM Difference between the 3 years buy-and-hold return (BHR) of the SEO firm and the Book-

to-Market and size 3 years matched portfolio BHR - - COMPUSTAT

3 years BHARs size Difference between 3 year BHR of the SEO firm and the 3 year BHR of the size matched firm - - COMPUSTAT

3 years BHARs value wieghted (v.w) market index

Difference between 3 y return of the SEO firm and the three-year BHR of the v. w market index. - - COMPUSTAT

37

Table II. 2

Frequency Distribution of Seasoned Equity Issues by Year and Industry

This table presents the distribution of our sample of 112 insured SEOs and 26 uninsured SEOs by fiscal year and industrial sector for the period from 1997 to 2003. Financial institutions are excluded form the sample. Other notable exclusions include IPOs, issues by non-Canadian firms, private placements, rights issues, and unit offerings.

Panel A. Distribution by year

Insured Firms Uninsured Firms

Year Number of firms % of sample Number of firms % of sample

1997 17 15.18 5 19.23

1998 14 12.50 4 15.38 1999 27 24.11 9 34.62 2000 27 24.11 3 11.54 2001 11 9.82 1 3.85 2002 8 7.14 3 11.54 2003 8 7.14 1 3.85 Total 112 100 26 100

Panel B. Distribution by Industry

Insured Firms Uninsured Firms

Industry Number of firms % of sample Number of firms % of sample

Agricultural industries 1 0.89 0 0.00 Biotechnology/pharmaceuticals 15 13.39 1 3.85 Communications and media 5 4.46 1 3.85 Construction 5 4.46 0 0.00 Consumer products 14 12.50 0 0.00 Industrial products 10 8.93 6 23.08 Merchandising 4 3.57 2 7.69 Mining and Natural Resources 20 17.86 12 46.15 Technology 26 23.21 4 15.38 Utility 3 2.68 0 0.00 Others 9 8.04 0 0.00

Total 112 100 26 100

38

Table II. 3

Descriptive Statistics This table presents the descriptive statistics of the sample of Canadian 112 insured SEOs and 26 uninsured SEOs for the period 1997-2003. Panel A reports the results for the D&O liability insurance variables, Panel B for the SEO characteristic variables, Panel C for the firm's business risk variables, Panel D for the corporate governance variables, and Panel E for the long-term performance measures. For each variable, we show the number of observation, the mean, the standard deviation, the median, the mean change, the median change, the t-statistic, and the Z-statistic.

Insured Firms (N=112) Uninsured Firms (N=26) Variable

N Mean St. Dev. Median N Mean St. Dev. Median

Panel A. D&O Liability Insurance Variables

Coverage ($millions) 112 18.506 13.804 15.000 26 Premium ($millions) 112 0.089 0.097 0.000 26 Coverage/SEO size 112 0.887 2.168 0.447 26 Premium/Coverage 112 0.005 0.005 0.004 26

Panel B. SEO Characteristics

SEO size ($millions) 112 52.313 51.165 33.442 26 63.147 151.149 24.700 Fraction sold percent 111 13.120 7.721 11.000 26 18.429 8.510 18.891

Panel C. Firm’s Business Risk Variables

Leverage 107 0.601 1.105 0.243 26 0.680 0.591 0.616 Age 112 15.717 15.126 11.000 26 27.100 62.397 13.250 Revenue ($millions) 112 372.831 1087.051 70.444 26 501.059 1365.791 17.967 SD adjusted revenue ($millions) 112 269.208 1772.923 34.343 26 0.423 0.504 23.665

U.S. exchange listing dummy 112 0.045 0.207 0.000 26 0.077 0.272 0.000

39

Table II. 3 - Continued Insured Firms (N=112) Uninsured Firms (N=26)

Variable N Mean St. Dev. Median N Mean St. Dev. Median

Panel D. Corporate Governance Variables

Board size 112 8.518 2.442 8.000 25 7.760 2.570 7.000 Pct. of outside directors 112 67.630 20.129 70.700 26 63.416 15.931 66.372 Blockholder 111 27.362 26.593 19.000 26 30.739 18.338 28.750 Executive cash comp ($millions) 112 1.814 1.614 1.363 26 1.389 1.468 1.008 D&O ownership (%) 112 13.516 16.693 6.495 26 14.499 18.543 10.592

Big 4 auditor 111 0.865 0.343 1.000 26 0.962 0.196 1.000 Panel E. Long-Run Performance Measures

3 years BHARs size & BM 112 -1.186 1.864 -0.907 26 -0.616 1.005 -0.529 3 years BHARs size 112 -0.892 1.759 -0.360 26 -0.779 1.608 -0.403 3 years BHARs value wieghted (v.w) market index 112 1.893 1.165 1.525 26 0.122 0.854 0.029

40

Table II. 4

Predicting the D&O Insurance Purchase Decision This table presents the results of a probit model designed to estimate the likelihood that an SEO purchase a D&O insurance policy. The explanatory variables are regressed on the dependent variable taking the value of 1 for firms with D&O insurance and 0 otherwise. The variables are defined in Table I. Coefficient estimates are presented with p-value in brackets below. Robust variance estimates are used. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***, respectively. Variable (1) (2) (3) (4)

1.579 1.134 1.612 1.208 Intercept (0.044)** (0.161) (0.047)** (0.188)

-0.012 -0.013 -0.013 -0.017 Age (0.008)*** (0.004)*** (0.007)*** (0.002)***

-0.110 -0.082 -0.114 -0.009 Leverage (0.325) (0.464) (0.313) (0.927)

0.001 0.001 0.001 0.001 Average revenue (0.032)** (0.030)** (0.026)** (0.234)

-0.001 -0.001 -0.001 0.000 Std dev revenue (0.021)** (0.020)** (0.017)** (0.111)

-0.165 -0.137 -0.173 0.077 U.S. exchange listing dummy (0.798) (0.833) (0.790) (0.409)

0.077 0.070 0.076 0.006 Board size (0.320) (0.351) (0.331) (0.477)

0.003 0.005 0.003 -0.007 Pct outside directors (0.745) (0.580) (0.744) (0.255)

-0.005 -0.005 -0.005 -0.003 Blockholder (0.364) (0.368) (0.359) (0.730)

-0.001 -0.002 -0.002 0.044 D&O ownership percent (0.881) (0.866) (0.871) (0.761)

0.097 0.111 0.098 -0.343 Executive cash compensation (0.388) (0.325) (0.388) (0.546)

-0.756 -0.590 -0.762 0.006 Big 4 auditor dummy (0.145) (0.232) (0.145) (0.176)

-0.003 -0.002 -0.003 -0.029 SEO Size (0.457) (0.664) (0.456) (0.129)

-0.027 -0.030 -0.027 -0.026 Fraction sold (0.100)* (0.098)* (0.100)* (0.163)

-0.205 3 y BHAR size and B/M (0.128)

0.015 3 y BHAR size (0.875)

-0.288 3 y BHAR v. w. Market index (0.275)

Number of observations 127 127 127 127 Adjusted R-squared 0.1474 17.28 14.76 24.57

41

Table II. 5 D&O insurance Coverage and Premium Variation Around SEO Announcement

This table presents the univariate results for the evolution from pre-SEO to post-SEO of D&O insurance coverage, premium, anddeductible around SEO date for the period 1997-2003. Panel A reports the mean before and after SEO, the change in the mean,and the p-value of the ttest for the difference in means. Panel B reports the median before and after SEO, the change in themedian, and the p-value of the Wilcoxon rank-sum test for the difference in medians. Significance at the 10 percent, 5 percent, and1 percent level is noted by *, **, and ***, respectively.

Variable N Before SEO After SEO Change p-value difference

Panel A. Mean

Coverage ($ millions) 111 13.938 19.080 5.142 0.000***

Premium ($ millions) 91 0.075 0.109 0.034 0.000***

Deductible ($ millions) 110 0.095 0.141 0.045 0.008***

Panel B. Median

Coverage ($ millions) 111 10 15 5 0.000***

Premium ($ millions) 91 0.039 0.047 0.008 0.000***

Deductible ($ millions) 110 0.025 0.05 0.025 0.000***

42

Table II. 6

Explaining the Total Amount of Coverage Purchased in the Secondary Market

This table shows the results of regressions that attempt to explain the amount of D&O insurance purchased by each SEO firm. The variables are defined in Table I. Coefficient estimates are presented with p-value in parentheses below. Robust variance estimates are used. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***, respectively.

Variable (1) (2) (3) (4)

Intercept 1.291 0.447 -0.763 5.591 (0.770) (0.918) (0.858) (0.252) Age 0.136 0.135 0.148 0.126 (0.017)** (0.018)** (0.006)*** (0.025)** Leverage 2.296 2.353 2.446 2.438 (0.042)** (0.034)** (0.036)** (0.027)** Average revenue 0.001 0.001 0.000 0.001 (0.834) (0.830) (0.865) (0.817) St. deviation revenue 0.000 0.000 0.000 0.000 (0.915) (0.931) (0.925) (0.832) U.S. exchange listing dummy 11.974 12.196 12.517 11.872 (0.000)*** (0.000)*** (0.000)*** (0.000)***Board size 0.440 0.437 0.499 0.260 (0.331) (0.329) (0.263) (0.570) Pct outside directors 0.042 0.045 0.038 0.044 (0.291) (0.250) (0.336) (0.285) Blockholder -0.111 -0.109 -0.103 -0.107 (0.010)*** (0.011)** (0.015)** (0.012)** D&O ownership percent -0.099 -0.101 -0.085 -0.120 (0.049)** (0.046)** (0.090)* (0.020)** Executive cash compensation 3.444 3.452 3.452 3.455 (0.000)*** (0.000)*** (0.000)*** (0.000)***Big 4 auditor dummy 6.040 6.279 6.052 6.250 (0.016)** (0.011)** (0.015)** (0.015)** SEO Size 0.055 0.058 0.054 0.050 (0.016)** (0.013)** (0.010)*** (0.029)** Fraction sold -0.317 -0.335 -0.302 -0.337 (0.004)*** (0.003)*** (0.006)*** (0.002)***3 y BHAR size and B/M -0.396 (0.100)* 3 y BHAR size -1.093 (0.004)***

3 y BHAR v. w. Market index -1.271 (0.020)** Number of observations 103 103 103 103 Adjusted R-squared 54.59 54.62 55.8 59.18

43

Table II. 7

Explaining the Insurance Premium

This table presents the results of regressions of second-stage regressions (equation (4)) that attempt toexplain the amount of premium paid by the SEO firm to purchase the D&O insurance policy. The first stage(equation (2)) is a regression of the total coverage purchased on all of the explanatory variables listed inTable V excluding variables unknown at the time of SEO. The residuals from that regression are labeled"Unexplained Coverage," and enter into the second stage regressions reported here. Variable definitions aregiven in Table I. Coefficient estimates are presented with p-value in parentheses below. Robust varianceestimates are used. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***,respectively.

Variable (1) (2) (3) (4)

Intercept 0.093 0.094 0.076 0.120

(0.160) (0.177) (0.197) (0.057)*

Age 0.001 0.001 0.001 0.001

(0.135) (0.139) (0.078)* (0.175)

Leverage 0.009 0.009 0.010 0.014

(0.314) (0.322) (0.273) (0.120)

Average revenue 0.000 0.000 0.000 0.000

(0.119) (0.118) (0.100)* (0.087)*

Std dev revenue 0.000 0.000 0.000 0.000

(0.111) (0.100)* (0.095)* (0.066)*

U.S. exchange listing dummy 0.088 0.087 0.090 0.089

(0.056)* (0.057)* (0.049)** (0.078)*

Board size 0.002 0.002 0.003 0.000

(0.541) (0.550) (0.476) (0.918)

Pct outside directors -0.001 -0.001 -0.001 -0.001

(0.200) (0.208) (0.191) (0.158)

Blockholder 0.000 0.000 0.000 0.000

(0.385) (0.389) (0.452) (0.381)

D&O ownership percent -0.001 -0.001 -0.001 -0.001

(0.183) (0.188) (0.261) (0.084)*

Executive compensation 0.018 0.018 0.017 0.018

(0.020)** (0.020)** (0.018)** (0.014)**

Big 4 auditor dummy -0.001 -0.002 0.000 -0.002

(0.960) (0.945) (0.993) (0.955)

SEO Size 0.000 0.000 0.000 0.000

(0.268) (0.301) (0.307) (0.655)

Fraction sold -0.002 -0.002 -0.002 -0.002

(0.021)** (0.026)** (0.036)** (0.013)**

44

Table II. 7 - Continued Variable (1) (2) (3) (4)

Unexplained coverage 0.004 0.004 0.003 0.015

(0.700) (0.697) (0.727) (0.237)

0.000 Unexplained coverage*BHAR Size & B/M

(0.878)

-0.003 Unexplained coverage*BHAR Size (0.081)*

-0.006 Unexplained coverage* v. w. market index

(0.056)* Number of observation 103 103 103 103

Adjusted R-squared 31.68 31.07 37.18 32.01

45

Table II. 8

Three-Stage Estimation of the Insurance Coverage and Premium This table reports the three-stage regression results of the relation between D&O coverage limit and premium for a sample of 112 Canadian insured SEOs for the period 1997-2003. Panel A examines the effect of long-term performance on the coverage level. Panel B investigates whether the D&O insurers are able to price the managerial opportunism translated into abnormal purchase of coverage. The variables are defined in Table I. Coefficient estimates are presented with p-value in parentheses below. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***, respectively.

Panel A. Explaining the Total Amount of

Coverage Purchased Panel B. Explaining the Insurance Premium

Variable (1) (2) (3) (1) (2) (3)

0.440 -0.983 5.930 0.077 0.060 0.102 Intercept (0.928) (0.836) (0.267) (0.123) (0.216) (0.039)** 0.135 0.149 0.125 0.001 0.001 0.001 Age

(0.013)** (0.005)*** (0.022)** (0.175) (0.116) (0.264) 2.354 2.462 2.449 0.002 0.003 0.007 Leverage

(0.002)*** (0.001)*** (0.001)*** (0.852) (0.745) (0.452) 0.001 0.000 0.001 0.000 0.000 0.000 Average revenue

(0.773) (0.830) (0.759) (0.146) (0.129) (0.127) 0.000 0.000 0.000 0.000 0.000 0.000 Std dev revenue

(0.914) (0.910) (0.785) (0.141) (0.124) (0.100)* 12.197 12.575 11.864 0.078 0.081 0.080 U.S. exchange listing dummy (0.001)*** (0.000)*** (0.001)*** (0.034)** (0.024)** (0.026)** 0.437 0.505 0.246 0.001 0.002 -0.001 Board size

(0.302) (0.221) (0.569) (0.789) (0.716) (0.901) 0.045 0.037 0.044 -0.001 -0.001 -0.001 Pct outside directors

(0.295) (0.374) (0.304) (0.133) (0.133) (0.100)* -0.109 -0.102 -0.107 -0.001 0.000 -0.001 Blockholder (0.001)*** (0.002)*** (0.001)*** (0.095)* (0.157) (0.098)* -0.101 -0.084 -0.122 -0.001 0.000 -0.001 D&O ownership percent (0.072)* (0.124) (0.031)** (0.304) (0.377) (0.100)*

3.452 3.453 3.456 0.014 0.014 0.015 Executive cash compensation (0.000)*** (0.000)*** (0.000)*** (0.015)** (0.016)** (0.009)***

6.280 6.053 6.267 0.002 0.004 0.002 Big 4 auditor dummy (0.016)** (0.016)** (0.014)** (0.949) (0.868) (0.926)

0.058 0.054 0.049 0.000 0.000 0.000 SEO Size (0.004)*** (0.005)*** (0.013)** (0.455) (0.467) (0.871)

-0.335 -0.300 -0.338 -0.002 -0.002 -0.002 Fraction sold (0.002)*** (0.003)*** (0.001)*** (0.048)** (0.043)** (0.020)**

3 y BHAR size and B/M -0.400 (0.336)

3 y BHAR size -1.209 (0.004)***

3 y BHAR v. w. Market index -1.371 (0.055)**

46

Table II. 8 - Continued

Panel B. Explaining the Total Amount of

Coverage Purchased Panel C. Explaining the Insurance Premium

Variable (1) (2) (3) (1) (2) (3)

0.016 0.015 0.026 Unexplained coverage (0.054)* (0.067)* (0.006)***

0.000 Unexplained coverage* 3 y BHAR size and B/M (0.842)

-0.003 Unexplained coverage* 3 y BHAR size (0.032)**

-0.006 Unexplained coverage* 3 y BHAR v. w. Market index (0.028)** Number of observations 103 103 103 103 103 103

Adjusted R-squared 65.08 66.88 65.76 30.17 33.38 33.43

Table II. 9

Explaining the Total Amount of Coverage Purchased by Multiple SEO Firms or Growth Firms

This table shows the results of regressions that attempt to explain the amount of D&O insurance purchased by each SEO firm. In panel A, dummy takes 1 for multiple SEOs and 0 otherwise, while in panel B, dummy takes 1 for growth firms and 0 otherwise. The variables are defined in Table I. Coefficient estimates are presented with p-value in parentheses below. Robust variance estimates are used. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***, respectively.

Multiple SEO Growth SEO Variable

(1) (2) (3) (4) (1) (2) (3) (4)

Intercept -3.877 -4.879 -5.395 4.462 1.652 1.049 1.416 2.810 (0.512) (0.423) (0.351) (0.369) (0.787) (0.865) (0.817) (0.644)

Age 0.104 0.104 0.111 0.069 0.085 0.090 0.085 0.084 (0.105) (0.110) (0.076)* (0.225) (0.197) (0.169) (0.197) (0.199)

Leverage 2.477 2.541 2.588 1.055 2.186 2.292 2.217 2.358 (0.055)* (0.046)** (0.049)** (0.100)* (0.078)* (0.063)* (0.077)* (0.050)**

Average revenue -0.001 -0.001 -0.001 0.003 -0.002 -0.002 -0.002 -0.002 (0.703) (0.729) (0.685) (0.337) (0.635) (0.631) (0.642) (0.625)

Std dev revenue 0.001 0.001 0.001 -0.001 0.001 0.001 0.001 0.002 (0.504) (0.532) (0.517) (0.636) (0.453) (0.453) (0.471) (0.376)

U.S. listing dummy 10.813 11.140 11.270 10.402 11.451 12.020 11.513 11.145 (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000***

Board size 0.707 0.696 0.765 0.332 0.663 0.682 0.661 0.473 (0.164) (0.162) (0.128) (0.506) (0.190) (0.170) (0.191) (0.362)

Pct outside directors 0.047 0.051 0.044 0.033 0.037 0.038 0.038 0.047 (0.263) (0.224) (0.285) (0.459) (0.382) (0.367) (0.368) (0.279)

Blockholder -0.078 -0.077 -0.070 -0.103 -0.083 -0.082 -0.082 -0.081 (0.177) (0.184) (0.220) (0.012)** (0.146) (0.152) (0.155) (0.160)

D&O ownership percent -0.091 -0.094 -0.080 -0.080 -0.108 -0.103 -0.106 -0.124 (0.110) (0.100)* (0.162) (0.145) (0.067)* (0.082)* (0.078)* (0.039)**

Executive cash compensation 3.125 3.143 3.117 2.419 3.110 3.127 3.111 3.170 (0.001)*** (0.001)*** (0.001)*** (0.000)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***

48

Table II. 9 - Continued Multiple SEO Growth SEO

Variable (1) (2) (3) (4) (1) (2) (3) (4)

Big 4 auditor dummy 7.541 7.825 7.454 5.338 7.311 7.543 7.385 7.877 (0.006)*** (0.004)*** (0.005)*** (0.022)** (0.008*** (0.006*** (0.008*** (0.005)***

SEO Size 0.080 0.083 0.078 0.075 0.077 0.079 0.078 0.073 (0.023)** (0.021)** (0.027)** (0.000)*** (0.023)** (0.021)** (0.023)** (0.038)**

Fraction sold -0.341 -0.363 -0.328 -0.246 -0.336 -0.364 -0.337 -0.360 (0.002)*** (0.001)*** (0.004)*** (0.009)*** (0.003)*** (0.002)*** (0.003)*** (0.002)***

Dummy 2.211 2.231 1.857 2.119 -3.003 -3.648 -3.292 -0.174 (0.253) (0.250) (0.341) (0.198) (0.208) (0.135) (0.192) (0.957)

-0.516 -0.645 3 y BHAR size &_B/M*dummy (0.084)* (0.036)**

3 y BHAR size*dummy -0.947 -0.380 (0.016)** (0.587)

-0.859 -1.585 3 y BHAR market index*dummy

(0.057)* (0.044)** Number of observations 103 103 103 103 103 103 103 103 Adjusted R-squared 54.71 54.700 55.720 59.450 54.87 55.01 54.48 55.4

Table II. 10 Explaining the Total Amount of Coverage Purchased in the Secondary

Market Using Heckman selection model

This table shows the results of the Heckman selection model that attempt to explain the amount of D&O insurance purchased by each SEO by controlling for a potential selection bias in the sample. The variables are defined in Table I. Coefficient estimates are presented with p-value in parentheses below. Robust variance estimates are used. Significance at the 10 percent, 5 percent, and 1 percent level is noted by *, **, and ***, respectively.

Variable (1) (2) (3) (4)

Intercept -2.616 -3.602 -4.415 0.483 (0.620) (0.505) (0.388) (0.938) Age 0.089 0.088 0.098 0.081 (0.144) (0.148) (0.093)* (0.169) Leverage 2.464 2.530 2.594 2.580 (0.034)** (0.026)** (0.028)** (0.022)**Average revenue -0.001 -0.001 -0.002 -0.002 (0.667) (0.692) (0.649) (0.653) St. deviation revenue 0.001 0.001 0.001 0.001 (0.470) (0.496) (0.482) (0.404) U.S. exchange listing dummy 10.521 10.843 11.040 10.391 (0.000)*** (0.000)*** (0.000)*** (0.000)***Board size 0.773 0.762 0.823 0.647 (0.100)* (0.098)* (0.078)* (0.193) Pct outside directors 0.051 0.055 0.047 0.052 (0.195) (0.160) (0.218) (0.189) Blockholder -0.074 -0.073 -0.067 -0.071 (0.170) (0.175) (0.214) (0.190) D&O ownership percent -0.093 -0.096 -0.082 -0.109 (0.074)* (0.066)* (0.116) (0.048)**Executive cash compensation 3.011 3.026 3.015 3.011 (0.001)*** (0.000)*** (0.000)*** (0.000)***Big 4 auditor dummy 7.103 7.384 7.102 7.296 (0.004)*** (0.003)*** (0.003)*** (0.003)***SEO Size 0.078 0.081 0.077 0.076 (0.016)** (0.014)** (0.017)** (0.023)**Fraction sold -0.343 -0.365 -0.329 -0.359 (0.001)*** (0.000)*** (0.001)*** (0.000)***3 y BHAR size and B/M -0.513 (0.065)* 3 y BHAR size -0.990 (0.005)***

3 y BHAR v. w. Market index -0.956 (0.100)* Number of observations 132 132 132 132